A used car repricing method based on K-Means++ clustering and multiple linear regression

A predictive pricing method based on k-means++ clustering and multiple linear regression is proposed to solve the unreasonable initial of used cars evaluate model's low accuracy. To begin, all vehicles are divided into three categories transaction period data sample. The K-Means++ algorithm then cluster best-selling cars, a equation fitted for each category as prediction evaluation. Finally, revaluation unsold unsalable examined. In addition, two car valuation methods, XGBoost AdaBoost, compared. Because model performs well, experiment shows that using estimate price reasonable.